Rakuten Institute of Technology at NLP 2019
The 25th annual meeting of the Association for Natural Language Processing, (NLP) was held at Nagoya University from March 12-15. The conference provides a leading forum for mutual exchange among researchers working in diverse fields, including linguistics, cognitive sciences, pedagogy and the humanities. RIT members were selected to sponsor a booth and hold three poster sessions at the 3-day conference.
In the first session, “Extracting Causes on Financial Performance from Company Financial Reports based on Sentence Cohesion”, researchers use an extraction method to understand and articulate the causes of financial performance from financial texts and reports. The method estimates the sentences that are most reliable and those without the cause, propagating estimation results to sentences with cohesion and showing their effectiveness. The second session, “Combination of Statistical and Neural Approaches in the Japanese Generation System” looks at approaches of simplifying the task of generating a natural question from a given input sentence. In so doing, the authors aim to establish a system that surpasses previous systems. Finally, “Design and Construction of Financial Statement Numeric Expression Corpus” targets securities reports, classifies account items of numerical expressions that appear in sentences and constructs a tag-assigning corpus. The purpose of the paper is to consider various numerical expressions in understanding the profitability and volatility of businesses.
We are grateful to have had the opportunity to share our findings with other leading scholars and researchers; more importantly, we are delighted to contribute to the conversation in the field of natural language processing.
We hope that we will be able to attend NLP in 2020.
For more information on NLP, please see the link below: